28/28 and, *most importantly*, thanks to all the domain experts who work on such important problems: they are those leading the impact, we are only providing tools. Let them be our inspirations! #useR2019
0
0
0
0
Hashtag
#useR2019
Advertisement · 728 × 90
aiforsocialgood.github.io/2018/pdfs/track1/80_aisg... decoders.azureedge.net/data-viz/images/Troll%20... https://openreview.net/forum?id=B1g-SnUaUN #useR2019
0
0
1
0
26/ Kudos to the amazing team making this possible every day: @saralouxcross @laure_delisle @alkalait Rafa Pardinas, Zhichao Lin, Michel Deudon, Jan Freiberg, Denis Kochetkov, @sherifea, Alex Kuefler, @BuffyPrice @ArchydeB @PhilBeaudoin and ... #useR2019
0
0
1
0
25/ Q: [How to find] ways to connect to give some virtual training to Africa in R and statistics.
A: Capacity development is brilliant. Check out @DeepIndaba and http://www.datascienceafrica.org/ leading the work there, or #MLSS2019.
That's all for #useR2019 Q&A, thanks to all!
0
0
1
0
24/ Q: Keep the internet anonymous, or keep the internet safe?
A: why can't we have both? Deferring to @ACLU and @AmnestyTech for better answers (@TanyaOCarroll @milena_iul ) , but physical space (streets) can be both mostly anonymous *and* safe, why not Internet? #useR2019
0
0
1
0
23/ ... Trying to create jobs at the interface (see @element_ai ).
Long term: Advocating for all companies to include this kind of work!
Longer term: Changing the incentives -- see @rcbregman brilliant book https://en.wikipedia.org/wiki/Utopia_for_Realists
#useR2019
0
0
1
0
22/ Q: How do you handle the temptation of earning a lot of money as a fulltime data scientist vs. Working for organizations which face heavy budget constraints?
A: Short term: Volunteering in addition to day job (see 11/).
Medium term: ..
#useR2019
0
0
1
0
21/ Q: Given that the situation in Sudan and Africa in general seems rather complicated, do you think you know enough to have a positive impact in the long run?
A: Me: no! @amnesty's brilliant folks: more than me :) Never a sure thing, but we try. @milena_iul, thoughts? #useR2019
0
0
1
0
20/ Q: Would you recommend one to do a PhD like yours?
A: What I do recommend is to learn to learn, and optimise for transferable/reusable skills. A PhD worked for me that way, but don't undertake one lightly, it can be brutal on mental health. #useR2019
0
0
1
0
19/ ... "effective altruism". My own choice: i/ taken to extreme: if we all work "for good" the world would be better, whereas if we all work for FANG we're doomed. ii/ we have only one life, what do you want to spend it on? #useR2019
2
0
1
0
18/ Q: Does the data scientist working "for good" have more impact than the data scientist working for FANG making six figures, donating to (possibly non-AI) causes?
A: *Excellent* question! Arguments pros and cons on both sides. Cf debate on ... #useR2019
0
0
1
0
17/ Q: What fonts were you using? Loved the talk (and the slides).
A: Haha thanks. Wish I could brag about my design skills, but it all comes from Keynote Template "Modern Type". DIN Condensed for titles, DIN Alternate for the rest. #useR2019
0
0
1
0
16/ Q: Was every tile uniquely tagged by one labeler or were there several for one tile involved?
A: Between 3 and 60 graders per tile. See full details in our @NeurIPSConf workshop article aiforsocialgood.github.io/2018/pdfs/track1/80_aisg... .#useR2019
0
0
1
0
15/ ... ESA challenge). https://dssg.uchicago.edu/ doing it brilliantly too. Many such initiatives! Great start. Scaling up sometimes needs more than students, though. #useR2019
0
0
1
0
14/ Q: How could you target universities with data science problems for the greater good? There you have workforce which is still not targeted for money.
A: Yoshua Bengio's @MILAMontreal and their #AIForHumanity group is such an example (we partnered on the ... #useR2019
0
0
1
0
13/ Q: Do you use far red or non-visible satellite imagery? Some features show up differently, like plants.
A: We want to! :) DecodeDarfur had only RGB available, Super-res is on NIR and Red channels. We are working on getting multispectral data.
#useR2019
0
0
1
0
0
0
1
0
11/ Q: What's the best way we contribute our skills for good?
A: Short-term: volunteer at @DataKind , https://correlaid.org/ @data4democracy , etc.
Medium-term: *advocate* to your boss to contribute your team's time to such efforts! Huge impact if we all do so. #useR2019
0
0
1
0
9/ Q: What other fields do you work on with philanthropic benefit? Do you also do work on microfinance or food security?
A: see 5/. We only started 18 months ago. Food security has deepest ramifications (@davidedjensen can say a lot about this), we'd love to work on it. #useR2019
0
0
1
0
8/ Q: Could have twitter started any repercussion after the publication of your study - or did it happen already? Does this affect the conception of such projects?
A: @milena_iul and @TanyaOCarroll can best answer. @jack met @kuminaidoo a few days after release. #useR2019
0
0
1
0
7/ ... (e.g. @amnesty, @hrw) and inviting speakers and visitors from multiple horizons to get ideas and viewpoints beyond those we can have on our own. Still a work in progress, and always will be! #useR2019
0
0
1
0
6/ Q: You spoke of the problems connected with «white savior». How do the companies you’ve worked with handled hiring people with varying understandings of «good»?
A: It's hard. Very international team! Surrounding ourselves with multiple domain experts ... #useR2019
0
0
1
0
5/ Q: In which other sectors do you provide support?
A: Currently working on Human Rights, and on Environment, but looking further. Basing ourselves on @UN Sustainable Development Goals #SDG. #useR2019
0
0
1
0
4/ Q: How do you choose your projects?
A: Mix of of technical feasibility, track record of impact by domain partner, and possibility to scale to reusable tools beyond one-off project. Will share full details later this year, as part of @UN https://aicommons.com/ #useR2019
0
0
1
0
3/ Q: When was the key turning point in your career where you switched to applications "for good"?
A: Probably when offered a Gandhi and a MLK Jr comic book biography age 7 :) Optimised early for versatility with CS and Math, to better focus on impact now.
#useR2019
0
0
1
0
2/ Q: Today's AI still need a lot of data, data that is mostly kept behind locked doors. How do you think we can solve this problem ?
A: 2-pronged: i/ tech with data efficiency (see super-res) and simulated training data, ii/ biz with consortiums and partnerships.
#useR2019
0
0
1
0
1/ Thanks to all #useR2019 attendees for the incredible reception of Friday's keynote https://youtu.be/G2SgOQv3PEM?t=3341 on @element_ai #AI4good work! I am both proud and humbled :) This thread answers the http://sli.do questions left unanswered - in 280 characters!?
0
0
1
0
0
0
0
0